Working with multi-dimensional images

import seaborn_image as isns
import numpy as np

from skimage import exposure # for adjusting image exposure
from skimage import filters  # for applying filters

Load a sample multi-dimension image array

cells = isns.load_image("cells")
cells.shape

(256, 256, 60)

Plot specific slices from 3-D data

g = isns.ImageGrid(cells, slices=[10,20,30,40,50,55])

You can apply changes to the entire grid. Here, we hide the colorbar and change the colormap.

g = isns.ImageGrid(cells, slices=[10,20,30,40,50,55], cbar=False, cmap="inferno")

Index along different axis

g = isns.ImageGrid(
    cells,
    slices=[10,20,30,40,50,55],
    axis=1,
    cbar=False
)

Variable step sizes and start/end points

g = isns.ImageGrid(
    cells,
    step=3,
    start=10,
    stop=40,
    cbar=False
)

Variable column length and figure size

Change column length and figure size using the col_wrap and height parameter, respectively

g = isns.ImageGrid(cells, cbar=False, height=1, col_wrap=10)

Map a function to transform the 3-D data

Here, we will adjust the exposure using the adjust_gamma function from scikit-image. This can done by passing the function object to the map_func parameter. Additional parameters to the function object can be passed as kwargs.

g = isns.ImageGrid(
    cells,
    map_func=exposure.adjust_gamma,
    map_func_kw={"gamma" : 0.5},
    cbar=False,
    height=1,
    col_wrap=10)

Plot 4d data

We now load a small sample of the cifar10 dataset.

This array will be composed of 60 images of size 32x32 with 3 channels, RGB

isns.set_image(origin="lower")

cifar10_sample = isns.load_image("cifar10")
cifar10_sample.shape

(60, 32, 32, 3)

Plot specific slices

g = isns.ImageGrid(cifar10_sample, slices=[4, 7, 9, 14, 19, 43])

This time images don’t have a color map, since the channels are always interpreted as RGB channels.

g = isns.ImageGrid(
    cifar10_sample,
    start=0,
    stop=12,
    step=2,
    axis=0
)

cifar10_sample = isns.load_image("cifar10")

# use sobel filter and adjust output datatypes
def sobel_3d(img, **extra_params):
    return np.asarray([filters.sobel(i, **extra_params)*256 for i in img], dtype=int)

g = isns.ImageGrid(
    cifar10_sample,
    map_func=sobel_3d,
    height=1.5,
    col_wrap=10,
)

Plot list of 3d images

cifar10_sample_list = isns.load_image("cifar10 list")
cifar10_sample_list[0].shape

(32, 32, 3)

g = isns.ImageGrid(cifar10_sample_list[:6])